Drew Smith Models 8 Neurochemicals to Drive Believable Robot Emotions
Updated
Updated · Hackaday · Jul 11
Drew Smith Models 8 Neurochemicals to Drive Believable Robot Emotions
1 articles · Updated · Hackaday · Jul 11
Summary
Kindalive uses eight simulated neurochemicals—including dopamine and cortisol—to generate a robot face’s emotional state from input text, aiming for more believable nonverbal responses.
The Python project avoids standard LLM sentiment labels and instead models chemical interplay and decay, letting short-term reactions and longer-term moods emerge dynamically.
Twelve Facial Action Coding System movements then translate that emotional mix into expressions on a dot-matrix face, with the same output adaptable to LED matrices or animatronic servos.
The approach targets a practical robotics problem: nonverbal cues often shape human communication more than words alone, and even simple, non-human faces can make machines easier to relate to.